Epidemiologic correlates of antibody response to human ... - CiteSeerX

0 downloads 0 Views 210KB Size Report
B Nonnenmacher MD PhD1, J Pintos PhD2, M C Bozzetti MD PhD1,. I Mielzinska-Lohnas5, A T LorinczPhD5, N Ikuta MD6,. G Schwartsmann MD PhD1, L L ...
Missing:
International Journal of STD & AIDS 2003; 14: 258–265

ORIGINAL RESEARCH ARTICLE

Epidemiologic correlates of antibody response to human papillomavirus among women at low risk of cervical cancer B Nonnenmacher MD PhD1, J Pintos PhD2, M C Bozzetti MD PhD1, I Mielzinska-Lohnas5, A T Lorincz PhD5, N Ikuta MD6, G Schwartsmann MD PhD1, L L Villa PhD4, J T Schiller PhD3 and E Franco PhD2 1Posgraduac ¸ a˜o

em Clõ ´nica Me´dica/UFRGS, Porto Alegre, Brazil, 2Departments of Epidemiology and Oncology, McGill University, Montreal, Canada, 3Laboratory of Cellular Oncology, NIH, Bethesda/MD, USA, 4Lugwig Institute for Cancer Research, Sa˜o Paulo, Brazil, 5Digene Company, Gaithsburg, MD, USA Summary: A population at low risk for developing cervical cancer in Southern Brazil was studied to identify the main determinants of serological response to human papillomavirus (HPV). Enzyme-linked immunosorbent assay tests were performed in 976 women to detect serum IgG antibodies against HPV 16 L1 virus-like particles (VLPs) and HPVs 16, 18, 6 and 11 L1 VLPs as a mixture of antigens. Women with four or more sexual partners were more likely to be seropositive than women with one partner (HPV 16 serology odds ratio [OR] ˆ 3.06, 95% con® dence interval [CI]: 2.0± 4.8; HPV 6/11/16/18 serology OR ˆ 4.64, 95% CI: 3.0± 7.2). HPV DNA and both serological responses were associated. Those positives to HPV 16 serology were twice as likely to have a cytological diagnosis of squamous intraepithelial lesions (SILs) than seronegatives (OR ˆ 2.07; 95% CI: 1.0± 4.5, and OR ˆ 1.73; 95% CI: 0.8± 3.8). Seropositivity to HPV 16 and HPV 6/11/16/ 18 antigens seem to be better markers of past sexual activity than current HPV infection, and humoral response to HPV 16 or HPV 6/11/16/18 may not be a strong indicator of cervical lesions in populations at low risk for cervical lesions. Keywords: human papillomavirus, ELISA serology, cervical cancer

Introduction During the last decades, several epidemiological studies have shown that markers of sexual activity, such as early age at ® rst intercourse and lifetime number of sexual partners, are the most important determinants of cervical cancer1,2. These ® ndings suggested that a sexually transmitted agent was the main cause of cervical cancer, and consensus review has asserted that this agent is the human papillomavirus (HPV) 3. There is abundant biological and epidemiological evidence to indicate that cervical infection by some HPV types, especially HPV 16, is the critical event in the genesis of cervical neoplasia4± 6. More than 90% of malignant and premalignant cervical lesions contain HPV DNA,

The sera examined in this study were collected under protocols that were approved by the HCPA ethical committee Correspondence to: Bernadete Nonnenmacher MD, Rua Comendador Rheingantz 431/401, Porto Alegre, RS, Brazil 90450-020 E-mail: 1) [email protected]; 2) [email protected]

reinforcing the role of HPVs in the induction of neoplasia4,7. In addition, a high proportion (15± 40%) of asymptomatic women have detectable cervical HPV infection 3,8, and the prevalence is higher among teenagers and young adults compared to older women. However, only a small percentage of those women will develop cervical cancer. Studies indicate that the risk of progression of a cervical lesion is signi® cantly higher among those patients infected with high-risk HPV types 9. Different studies have demonstrated that HPV’s structural L1 protein alone, or L1 with L2 protein self-assembles into virus-like particles (VLPs) that can be used as antigens for the detection of type restricted antibodies to HPV16, or other types, in enzyme-linked immunosorbent assays (ELISAs)10± 17. We used ELISA tests with HPV 16 VLPs as single antigen, and HPV 16, 18, 6 and 11 VLPs as a mixture of antigens, to determine HPV seropositivity in a population of sexually active women who attended an outpatient clinic in the city of Porto Alegre, Southern Brazil. We sought to determine the epidemiological predictors of the humoral response

258

Nonnenmacher et al. Epidemiologic correlates to HPV antibody response

against HPV and the relationship between serology and presence of cervical lesions.

Patients and methods Study design The study followed a cross-sectional design. All women who attended the outpatient clinic at the `Liga Feminina de Combate ao CaÃncer’, in Porto Alegre, Southern Brazil, during July and August of 1994 were invited to participate. Women with a history of hysterectomy, conization, or cervical cancer were considered not eligible. Of the 1035 eligible women contacted, 977 consented to participate, and their ages ranged from 15 to 70 years. We collected information on socioeconomic, demographic, and reproductive history using a questionnaire-based interview done by a research nurse trained for this study. Variables known or suspected to be related to cervical cancer and HPV infection were included in the questionnaire, such as age, ethnicity, education, income, smoking habits, marital status, parity, use of contraceptive methods, previous abnormal Papanicolaou (Pap) test, lifetime and recent number of sexual partners, age at ® rst sexual intercourse, and previous gynaecological infection.

Cervical specimens An endocervical brush was used to collect cervical cells to prepare Pap smears for cytopathological diagnosis. Two different pathologists, blinded to the clinical and colposcopic diagnoses, examined the smears. The Bethesda International Classi® cation18 was used to assess cytological results. Cells of the squamous-columnar junction were also collected with the endocervical brush, and they were stored in phosphate buffered saline (PBS) solution at - 808C for subsequent genital HPV DNA analysis. All samples were tested for viral DNA using two methods: the MY09/11 consensus primer polymerase chain reaction (PCR) protocol19, and the hybrid capture II assay20. Consensus-PCR reactions were run with DNA samples from HeLa and SCC (CaSki) cell lines with/without HPV infection as controls for HPV typing. In addition, the b-globin gene was ampli® ed as an internal control. The use of internal controls during the PCR reaction con® rms that our conditions are working well, and therefore, we feel con® dent of successfully typing for HPV on the remaining samples. The hybrid capture assay includes a mixture of probes for the following cervical cancer-associated HPV types: 16, 18, 31, 33, 35, 39, 35, 51, 52, 56, 58, 59 and 68. The US Food and Drug Administrationapproved threshold of 1 pg of HPV DNA/mL of test solution was used for a positive result. An aliquot of specimen transport medium was denatured to produce single-stranded DNA and

259

reacted with a cocktail of the 13 full-length RNA probes recognizing oncogenic HPV types. Hybrids consisting of target HPV bound to RNA probes were bound or `captured’ on sides of tubes coated with antibodies recognizing DNA:RNA hybrids. Adding a second antibody tagged with alkaline phosphatase permitted detection of bound hybrid by a chemiluminescent readout. Test specimens in which light emission (expressed as relative light units [RLUs]) equalled or exceeded the mean of positive controls (PCs) consisting of 10 pg/mL HPV 16 DNA run in replicate were categorized as positive (RLU/CO51.0). In this study, the assay was qualitative so every sample with RLU/CO ratio 1 or above was considered positive20.

ELISA technique Blood specimens were collected by venipuncture. The blood specimens were centrifuged and the serum samples were stored at - 808C until tested. Each serum sample was tested for reactivity to HPV16 VJPs and to a mixture of VLPs from HPVs 16, 18, 6 and 11. The VLP-ELISA technique has been previously described in detail10. Brie¯ y, L1 VLPs were puri® ed by caesium chloride gradient centrifugation from cell lysate of Sf-9 cells that had previously been infected with recombinant baculovirus vectors expressing L1 from one of the four HPV types. Ninety-six-well polystyrene plates were coated with VLP particles (aliquots containing 1.0 mg) in 50 mL of PBS followed by incubation at 378C for 1.5 hours, and washing three times in calcium- and magnesium-free PBS. After a two-hour blocking step with PBS containing 0.5% non-fat dry milk and 0.1% newborn calf serum with three subsequent washes, 10 mL of serum diluted in 40 mL of 0.5% dry milk in PBS were added to the wells. The plates were incubated for 2.5 hours and washed ® ve times, after which horseradish peroxidaseconjugated goat F(ab0 )2 anti-human IgG diluted 1:10,000 in 0.5% milk-PBS was added. After onehour incubation at room temperature and three washes, 50 mL of substrate were added to each well with a 45-minute incubation at room temperature. A microplate reader was used to read optical densities at 405 nm. Serum specimens were assayed in triplicate14. In order to control the variation of measured reactivity that is inherent to immunoenzymatic techniques we included in every ELISA plate a human serum pool in triplicate diluted in the same way as the specimens. This serum pool was prepared beforehand and kept frozen at - 208C. It included dozens of blood bank and normal biochemistry lab specimens from adult donors. The same serum pool batch was used throughout the study. The uncorrected serological reactivity was measured as net absorbance values after subtracting from each specimen’s average absorbance value the mean blank reactivity from triplicate wells exposed only to diluent instead of diluted

260

International Journal of STD & AIDS Volume 14 April 2003

serum. Such `noise’ subtracted reactivity was further corrected by dividing each specimen’s mean value by the equivalent result from the serum pool included in the same plate in triplicate. The reactivity of this serum pool control varied between plates and between ELISA runs as a consequence of the vagaries of a complex technique such as the VLP-ELISA, which is prone to be affected by the variation in reactivity of different batches of reagents (antigen, conjugate, and chromogen-substrate mixture) and betweenplate and between-run variation due to the ¯ uctuation in ef® ciency of washing and binding properties of the polystyrene. By taking the ratio of the serological reactivity of each specimen to that of the serum pool from the same plate, we corrected for the net in¯ uence of these various reaction parameters, thus reducing the effect of inter- and intra-assay variations. Absorbance ratios represent multiples of reactivity with respect to that of the serum pool, which presumably re¯ ects the average antibody response from many donors of comparable age.

DeŽnition of serological positivity ELISA absorbance values, either uncorrected or corrected, are continuously distributed. Therefore, to classify samples as either negative or positive for anti-HPV antibodies it was necessary to apply an arbitrary cutoff point to de® ne seropositivity for HPV 16 VLPs, and for the mixture of HPVs 16, 18, 6 and 11 VLPs. The cutoff points were determined based on the percentiles of the distribution of absorbance ratios of a subgroup of women at low risk to develop genital HPV infection. This was de® ned as those women with age at ® rst sexual intercourse at 19 years or older, with only one lifetime sexual partner, without cytological abnormalities, and whose cervical samples were negative for HPV DNA testing by both PCR and hybrid capture. A total of 171 women met this pro® le. We then estimated for the entire patient population the odds ratios (ORs) for cervical HPV infection, as determined by the PCR technique, using as serology cutpoints the absorbance ratios corresponding to percentiles in the 30th± 90th range as determined in the low-risk sample. We chose as ® nal cutpoint for each antigen preparation an absorbance ratio in the region in which ORs were maximal within the latter range. Based on these two empirically-de® ned cutpoints (one for HPV-16 and one for the mixed-type ELISA), the serological results were classi® ed as negative and positive for each antigenic preparation, and interpreted as such for the analysis of serological response to HPV as a dichotomous outcome.

Statistical analysis To examine the main predictors of serological response anti-HPV, two binary outcomes were

considered: seroreactivity to HPV 16 VLPs, and seroreactivity to the combined HPVs 16, 18, 6, 11 VLPs. Unconditional logistic regression was used to estimate ORs and their 95% con® dence intervals (95% CIs) as the measure of magnitude of association between predictors and serological response. Age and presence of cervical HPV DNA by PCR were controlled for in the models. The ® nal model to determine the main predictors of serological response was ® tted using the stepwise backwards method (P-value for removalˆ 0.15, P-value for entry ˆ 0.10). We also estimated the association between HPV serology and cervical lesions using unconditional logistic regression. Serological response was also used as a binary predictor, while outcomes were de® ned as cytological or histological diagnosis of squamous intraepithelial lesions (SILs) of either low (LGSIL) or high grade (HGSIL). Diagnoses of `atypical squamous cells of undetermined signi® cance’ (ASCUS) and non-squamous abnormalities were excluded from this analysis. To con® rm the choice of serological cutpoints as de® ned above for HPV DNA positivity as an outcome, we also evaluated the combined diagnostic accuracy to detect SILs (estimated by sensitivity plus speci® city) as a function of distinct percentile cutpoints.

Results A total of 976 women were included in this study. Most women were married (74.2%), in the age group of 35 to 49 years (46.5%), white ethnicity (85%), with an elementary or lower education level (70.6%), and never smokers (56%). Regarding their reproductive life, 33.9% of them had had three or more children. A signi® cant proportion did not use any contraceptive methods during the six months previous to the interview (34.3%), or used oral contraceptives (32.5%). Eighty-one per cent of the patients reported that they had never had an abnormal Pap test. Most women (51.5%) had started their sexual life before 18 years of age, and had had a single lifetime sexual partner (55.7%). Further, 91.4% of the participating women denied any gynaecological infection in the six months previous to the interview. Figure 1 shows the ORs for HPV infection according to the percentile cutpoints for the HPV 16 and the HPV 6/11/16/18 antigen serology. The serological cutpoint that maximized the risk of being HPV positive by PCR was located between the 60th and the 80th percentile (Figure 1, bottom). Based on these results, we arbitrarily determined the cutpoints as the 75th percentile of the distribution of absorbance ratios for the subset of women considered to be at low risk to acquire genital HPV infection. Using these de® ned cutpoints Ð absorbance ratios of 0.678 for HPV 16 and 0.742 for the HPV 6/11/16/18 antigen Ð a total of 378 samples (38.7%) were positive for the HPV 16 antigen, and

Nonnenmacher et al. Epidemiologic correlates to HPV antibody response

Figure 1. Association between serological response and HPV DNA positivity by polymerase chain reaction (top graph) and diagnostic accuracy of HPV serology to detect cervical lesions (bottom graph) according to percentiles of the distribution of cutpoints used to define seroreactivity among the subset of women at low risk of cervical HPV disease (see text for details). Top: Odds ratios of HPV DNA positivity; bottom: combined sernsitivity and specificity of serology to detect squamous intraepithelial lesion Solid line ˆHPV 16 antigen, broken lineˆ HPV 6/11/16/18 antigen

382 (39.1%) were positive for the HPV 6/11/16/18 antigen. Table 1 shows the association between sociodemographic characteristics and seropositivity for both antigens. Women aged 35 to 49 years had a higher risk of being seropositive for the HPV 16 and the HPV 6/11/16/18 antigen compared to women under 25. Women 50 years or older had a signi® cantly higher risk of being seropositive for the HPV 6/11/16/18 antigen (OR ˆ 1.66; 95% CI: 1.0± 2.7), but not for HPV 16, when controlling for HPV DNA positivity. Married women were less likely to be seropositive for both antigens, compared to single women. Women of non-white ethnicity had a slightly higher risk of being seropositive, but the association was not signi® cant. Likewise, Table 2 presents the results for sexual behaviour variables and reproductive history as predictors of seropositivity. As expected, a strong positive association was found between lifetime number of sexual partners and seropositivity. Women with four or more partners were three to four times more likely to be seropositive than women with one sexual partner. The association

261

was stronger for the mixed-antigen ELISA (ORˆ 4.5, 95% CI: 2.9± 7.0) than for HPV 16 (OR ˆ 2.9; 95% CI: 1.9± 4.5). Number of partners in the six months previous to the interview was not as strongly associated with seroreactivity. Age at ® rst sexual intercourse was associated with both antigens with a signi® cant negative trend in positivity as age at ® rst encounter increased. Women with ® rst intercourse at 22 or older had almost half the risk of being seropositive than women with ® rst intercourse at 16 or younger: ORˆ 0.55; 95% CI: 0.4± 0.8 for HPV 16, and OR ˆ 0.58; 95% CI: 0.4± 0.9 for the mixed HPV antigen. Although parity was not consistently correlated with seroreactivity, nulliparous women with antecedents of miscarriage or abortion had an increased risk of seropositivity for the HPV 6/11/16/18 antigen, but not for HPV 16, compared to nulliparous women without such histories. Table 3 shows the analysis of HPV DNA in cervical samples as predictor of seropositivity. HPV DNA was detected in 14.6% of the samples using the PCR technique, in 13.6% using hybrid capture, and 23.2% using either diagnostic method. The association with the hybrid capture method was stronger for the HPV 16 antigen, whereas detection of HPV DNA by PCR was more strongly associated with the HPV 6/11/16/18 antigen. Table 4 shows models containing the most signi® cant independent predictors of serological response by antigenic preparation. The ® nal model for seroreactivity anti-HPV 16 as an outcome included age, smoking, parity, lifetime number of sexual partners, and detection of HPV DNA. The model for HPV 6/11/16/18 antigens included as predictors smoking, lifetime number of sexual partners, contraception methods, and detection of HPV DNA. The strongest predictor in both models was number of sexual partners, being the association slightly stronger for the HPV 6/11/16/18 antigen. Of the 976 Pap smears, 88.7% were diagnosed as within normal limits, 6.3% as ASCUS, 2.1% as LGSIL, and 0.6% as HGSIL. Among the 14.3% women who underwent a cervical biopsy only 3.1% were con® rmed histologically as having SILs: 2.7% with LGSIL and 0.4% with HGSIL. Table 5 shows the analysis where serological response is studied as a predictor of SILs de® ned either cytologically or histologically. The strongest association, although not signi® cant, was found between anti-HPV 16 seroreactivity SILs by cytology. All associations became attenuated after incremental adjustment for HPV DNA.

Discussion The characteristics of the study participants are consistent with a population with a low risk pro® le for genital HPV infection and cervical cancer. Recruitment took place in an outpatient clinic offering a screening programme for prevention of

262

International Journal of STD & AIDS Volume 14 April 2003 Table 1. Odds ratios (ORs) and 95% confidence intervals (CIs) for seropositivity according to sociodemographic and related variables HPV 16 seropositivity

HPV 6/11/16/18 seropositivity

Age‡HPV DNA adjusted OR (95% CI)

%

Age‡HPV DNA adjusted OR (95% CI)

Variable

Categories

n

%

Age (years)

424 25–34 35–49 550

113 232 454 175

30.1 38.8 42.4 35.4

1.0 1.56 (1.0–2.5) 1.87 (1.2–2.9) 1.37 (0.8–2.3)

32.7 38.4 39.9 42.3

1.0 1.38 (0.9–2.2) 1.52 (1.0–2.4) 1.66 (1.0–2.7)

Race

White Non-white

830 134

37.9 44.0

1.0 1.26 (0.9–1.8)

38.6 42.5

1.0 1.17 (0.8–1.7)

Schooling level

5Elementary Elementary High School Postsecondary

473 205 225 69

38.3 35.1 44.4 36.2

1.0 0.87 (0.6–1.2) 1.25 (0.9–1.7) 0.90 (0.5–1.5)

38.5 34.6 45.3 37.7

1.0 0.85 (0.6–1.2) 1.29 (0.9–1.8) 0.94 (0.6–1.6)

Income

53 minimum wages 43 minimum wages

503 392

38.2 39.6

1.0 1.04 (0.8–1.4)

38.6 39.5

1.0 1.05 (0.8–1.4)

Smoking

Never smoked Current smoker Former smoker

547 283 140

38.8 38.2 38.8

1.0 0.94 (0.7–1.3) 0.99 (0.7–1.5)

40.2 37.5 38.6

1.0 0.87 (0.6–1.2) 0.89 (0.6–1.3)

Marital status

Single Married Divorced Widowed

114 725 83 51

44.7 36.9 48.2 37.3

1.0 0.66 (0.4–1.0) 1.05 (0.6–1.9) 0.69 (0.3–1.5)

47.4 36.7 43.4 49.0

1.0 0.61 (0.4–0.09) 0.78 (0.4–1.4) 0.89 (0.4–1.8)

Table 2. Odds ratios (ORs) and 95% confidence intervals (CIs) for seropositivity according to reproductive and sexual behaviour variables HPV 16 seropositivity %

HPV 6/11/16/18 seropositivity

Age‡HPV DNA adjusted OR (95% CI)

%

Age‡HPV DNA adjusted OR (95% CI)

Variable

Categories

n

Parity

Nulliparous 1 2 3 or ‡ Miscarriage

143 197 262 331 40

33.6 39.6 32.4 44.2 50.5

1.0 1.27 0.90 1.48 2.06

(0.8–2.0) (0.6–1.4) (0.9–2.3) (1.0–4.3)

39.2 38.6 35.5 42.0 42.5

1.0 0.99 0.82 1.06 1.22

Contraception last six months

None OC use Tubal ligation Other

335 318 181 138

39.1 34.6 43.9 42.0

1.0 0.76 (0.5–1.1) 1.13 (0.8–1.7) 1.01 (0.7–1.6)

40.0 34.6 43.6 42.8

1.0 0.85 (0.6–1.3) 1.26 (0.9–1.9) 1.18 (0.8–1.8)

History of abnormal Pap

No Yes

786 180

39.0 37.8

1.0 0.98 (0.7–1.3)

39.8 36.1

1.0 0.87 (0.6–1.2)

Lifetime no. sex partners

1 2 3 54

544 205 114 110

30.9 45.4 45.6 57.3

1.0 1.89 (1.4–2.6) 1.82 (1.2–2.8) 2.95 (1.9–4.5)

27.9 49.8 49.1 63.6

1.0 2.55 (1.8–3.6) 2.47 (1.6–3.8) 4.53 (2.9–7.0)

No. partners last six months

0 or 1 2 or 3

954 20

38.6 50.0

1.0 1.63 (0.7–4.0)

38.9 55.0

1.0 2.00 (0.8–4.9)

Age at Žrst sexual intercourse

416 17–18 19–21 522

230 274 261 209

43.0 43.4 35.0 32.5

1.0 0.96 (0.7–1.4) 0.63 (0.4–0.9) 0.55 (0.4–0.8)

43.5 40.5 37.5 34.4

1.0 0.84 (0.6–1.2) 0.71 (0.5–1.0) 0.58 (0.4–0.9)

Genital infection last 6 months No HPV Other

893 16 46

38.4 33.3 50.0

1.0 0.78 (0.3–2.4) 1.63 (0.9–3.0)

39.1 31.3 43.5

1.0 0.66 (0.2–2.0) 1.23 (0.7–0.23)

OC ˆ oral contraceptive

(0.6–1.6) (0.5–1.3) (0.7–1.7) (0.6–2.5)

Nonnenmacher et al. Epidemiologic correlates to HPV antibody response

263

Table 3. Odds ratios (ORs) and 95% confidence intervals (CIs) for seropositivity according to detection of HPV DNA HPV 16 seropositivity

HPV 6/11/16/18 seropositivity

%

%

Age-adjusted OR (95% CI)

Age-adjusted OR (95% CI)

Detection technique

Capture

n

Hybrid II

Negative Positive

731 133

36.1 47.4

1.0 1.71 (1.2–2.5)

37.7 45.9

1.0 1.45 (1.0–2.1)

Polymerase chain reaction

Negative Positive

757 143

36.6 47.6

1.0 1.73 (1.2–2.5)

36.3 52.4

1.0 2.04 (1.4–2.9)

Either

Negative Positive

606 227

35.3 44.9

1.0 1.61 (1.2–2.2)

36.6 46.3

1.0 1.55 (1.1–2.1)

Table 4. Odds ratios for the most predictive variables* of the anti-HPV seroreactivity according to the type of antigen used in the enzyme-linked immunosorbent assay reaction Odds ratio (95% CI) HPV 16 antigen

HPV 6/11/16/18 antigen

Variable

Categories compared

Age (years)

25–34 vs. 525 35–49 vs. 525 50‡ vs. 525

1.64 (0.8–2.3) 1.66 (1.0–2.8) 1.17 (0.7–2.1)

Smoking

Current smoker vs. never Former smoker vs. never

0.78 (0.6–1.1) 0.86 (0.6–1.3)

Parity

1 vs. nulliparous 2 vs. nulliparous 53 vs. nulliparous Miscarriage vs. nulliparous

1.37 1.16 1.81 1.79

Lifetime sexual partners

2 vs. 1 3 vs. 1 54 vs. 1

1.97 (1.4–2.8) 1.96 (1.3–3.0) 3.06 (2.0–4.8)

2.66 (1.9–3.7) 2.56 (1.7–3.9) 4.64 (3.0–7.2)

HPV polymerase chain reaction

positive vs. negative

1.58 (1.1–2.3)

1.77 (1.2–2.6)

Contraception

Oral contraceptives vs. none Tubal ligation vs. none Others vs. none

0.70 (0.5–1.0) 0.74 (0.5–1.1)

(0.8–2.2) (0.7–1.9) (1.1–2.9) (0.8–3.8)

0.79 (0.6–1.1) 1.32 (0.9–2.0) 1.19 (0.8–1.8)

*By stepwise multiple logistic regression (see Methods for details). CIˆ conŽdence interval

Table 5. Odds ratios (ORs) and 95% confidence intervals (CIs) for cervical intraepithelial lesions by anti-HPV serological response and diagnostic approach to define lesions Age-adjusted OR (95% CI)

OR adjusted for age and HPV DNA (95% CI)

Lesion deŽnition*

HPV antigen types

Cytological{

HPV 16 HPV 6/11/16/18

2.07 (1.0–4.5) 1.50 (0.7–3.2)

1.73 (0.8–3.8) 1.16 (0.5–2.6)

Histological{

HPV 16 HPV 6/11/16/18

1.13 (0.5–2.6) 1.21 (0.5–2.9)

1.10 (0.5–2.6) 1.14 (0.5–2.9)

*Cytology: comparison between any grade squamous intraepithelial lesions (SILs) and normal cytological results (atypical squamous cells of undetermined signiŽcance and other categories excluded). Histology: comparison between conŽrmed any grade SILs and negative results among women who underwent biopsy { 894 women included in the analysis, 27 of them presented lesions { 133 women included in the analysis, 30 of them presented lesions

cervical cancer. Most women attending this clinic are asymptomatic, and they are concerned with cancer prevention. The majority of the women included in our study have had only one lifetime sexual partner.

In the evaluation of the ELISA results, we controlled for intra- and inter-assay ¯ uctuations by including a control serum pool sample. Previous studies have also used the mean of multiple measures corrected for control sera, and the cutoff point

264

International Journal of STD &AIDS Volume 14 April 2003

for de® nition of seropositivity was based on serological values of women at low risk to develop HPV infection12± 16. In our study, the analysis of determinants of seropositivity for both types of antigens showed that the strongest predictor was lifetime number of sexual partners. This ® nding is in agreement with results from previous studies12,14± 17, 21,22. Age at ® rst sexual intercourse, another marker of sexual activity used in epidemiological studies of cervical cancer, was negatively correlated with seropositivity. These results suggest that in this study population serology is primarily a marker of past HPV infection. Another factor associated with seropositivity for the HPV 6/11/16/18 antigen was multiparity. It is conceivable that every pregnancy increases the chance of acquiring an HPV infection, or exacerbates an already existing infection. Although nonwhite ethnicity has been shown to be linked to HPV infection and cervical neoplasia23, we did not ® nd a signi® cant association with seropositivity. A paradoxical ® nding in our study is that current and former smokers had lower levels of seropositivity for both the antigens than never smokers. Tobacco smoking reduces the immune response, but on the other hand seems to be a correlate of HPV infection24. The analysis of detection of HPV DNA showed signi® cant associations between both detection techniques and both serological responses analysed, although the magnitude of the associations were not as strong as the ones seen in the analysis of number of sexual partners. While detection of HPV DNA using the PCR technique was a better predictor for HPV 6/11/16/18 antigen reactivity, the hybrid capture technique was a better predictor for the HPV 16 serology. These results could be tentatively explained by the fact that the latter method is designed to detect high risk HPV types, while the MY09/11 PCR protocol detects a large spectrum of both low and high risk HPV genital types. The fact that HPV seropositivity was more strongly associated with markers of sexual activity than with detection of cervical HPV DNA suggests that serological response is more a marker of past rather than current HPV infection in this population. These results are in agreement with the ® ndings of a previous study in a population with a high risk for developing cervical cancer in Greenland16, where seropositivity was associated with number of sexual partners but not with detection of HPV DNA. The participants in our study were relatively old, compared with those of other studies in which a stronger association between VLP serology and concomitant genital HPV was seen 12,21. Together, these ® ndings reinforce the concept that most genital tract HPV infections in women are transient, but most seroconversions are persistent. Previous investigations have found strong associations between serological response to HPV-16 and HGSIL13,25,26. In the present study, we found a

higher risk of SIL for seropositivity to HPV 16 than for the HPV 6/11/16/18 antigen. The results were not signi® cant, but the estimates were very imprecise due to the small number of LGSIL and HGSIL diagnosed among the participants. It is unlikely that the low magnitude of the ORs could be attributable to the wrong choice of cutpoint to determine seroreactivity. The chosen 75th percentiles were within the range that yielded the strongest association with HPV DNA positivity as well as the most discriminating combination of sensitivity and speci® city for detecting cervical disease. Our ® ndings suggest that HPV serology is more a marker of past HPV infection rather than a marker of cervical lesions or current HPV infection. Our results support the hypothesis that either HPV 16 or HPV 6/11/16/18 VLP serological assays are not good indicators of women infected by HPV 6, 11, 16 or 18 at increased risk of cervical cancer in low risk cohorts of the type studied here. Acknowledgements: We thank Drs Joao Carlos Prolla and AntoÃnio Carlos PuÈtten for the cytology review. Financial support: FAPERGS and CAPES (Brazil).

References 1

Brinton LA. Epidemiology of cervical cancer overview. IARC Sci Publ 1992;119:3± 23 2 Franco OEL. Human papillomavirus and the natural history of cervical cancer. Infect Med 1993;10:57± 64 3 International Agency for Research on Cancer. IARC monographs on the evaluation of carcinogenic risks of chemicals to humans. Human papillomaviruses Vol 64, International Agency for Research on Cancer, Lyon, France, 1995 4 Zur Hausen H. Molecular pathogenesis of cancer of the cervix and its causation by speci® c human papillomavirus types. Curr Top 1994;186:131± 56 5 Schiffman MH. Latest HPV ® ndings Ð some clinical implications. Contemporary Obstet/Gynecol. 1993:27± 37 6 Franco EL. Cancer causes revisited Ð human papillomavirus and the cervical neoplasia. J Natl Cancer Inst 1995;87: 779± 80 7 Bosch FX, Manos MM, Munoz N, et al. Prevalence of human papillomavirus in cervical cancer: a worldwide perspective. J Natl Cancer Inst 1995;87:796± 802 8 Schiffman M, Bauer HM, Hoover RN, et al. Epidemiologic evidence showing that human papillomavirus infection causes most cervical intraepithelial neoplasia. J Natl Cancer Inst 1993;85:958± 64 9 Schiffman MH. Recent progress in de® ning the epidemiology of human papillomavirus infection and cervical neoplasia. J Natl Cancer Inst 1992;84:394± 8 10 Kirnbauer R, Hubbert NL, Wheeler CM, Becker TB, Lowy DR, Schiller JT. A virus-like particle enzyme-linked immunosorbent assay detects serum antibodies in a majority of women infected with human papillomavirus type 16. J Natl Cancer Inst 1994;86:494± 504 11 Galloway D. Papillomavirus capsids: a new approach to identify serologic markers of HPV infection. J Natl Cancer Inst 1994;86:474± 5 12 Dillner L, Wiklund F, Lenner P, et al. Antibodies against linear and conformational epitopes of human papillomavirus type 16 that independently associate with incident cervical cancer. Int J Cancer 1995;60:377± 82

Nonnenmacher et al. Epidemiologic correlates to HPV antibody response 13

14

15

16

17

18

19

Nonnenmacher B, Hubbert NL, Kirnbauer R, et al. Serologic response to human papillomavirus type 16 (HPV-16) viruslike particles in HPV-16 DNA Ð positive invasive cervical cancer and cervical intraepithelial neoplasia grade III patients and controls from Colombia and Spain. J Infect Dis 1995;172:19± 24 Wideroff L, Schiffman MH, Nonnenmacher B, et al. Evaluation of HPV16 virus-like particles seropositivity in an incident case± control study of cervical neoplasia. J Infect Dis 1995;172:1425± 30 Wideroff L, Schiffman MH, Hoover R, et al. Epidemiologic determinants of seroreactivity to human papillomavirus (HPV) type 16 virus like-particles in cervical HPV-16 DNApositive and negative women. J Infect Dis 1996;174:937± 43 Nonnenmacher B, Kruger Kjaer S, Svare EI, et al. Seroreactivity to HPV16 virus-like particles as a marker for cervical cancer risk in high-risk populations. Int J Cancer 1996;68:704± 9 Svare EI, Kjaer SK, Nonnenmacher B, et al. Seroreactivity to human papillomavirus type 16 virus-like particles is lower in high-risk men than in high-risk women. J Infect Dis 1997; 176:876± 83 National Cancer Institute Workshop: The 1988 Bethesda System for reporting cervical/vaginal cytologic diagnosis. Acta Cytol 1989;33:567± 74 Bernard HU, Chan SY, Manos MM, et al. Identi® cation and assessment of known and novel human papillomavirus by polymerase chain reaction ampli® cation, restriction fragment length polymorphism, nucleotide sequence, and phylogenetic algorithms. J Infect Dis 1994;170:1011± 85

20

21

22

23

24

25

26

27

265

Lorincz AT. Hybrid capture method for detection of human papillomavirus DNA in clinical specimens. Papillomavirus report 1996;7:1± 5 Carter JJ, Koutsky LA, Wipf GC, et al. The natural history of human papillomavirus type 16 capsid antibodies among a cohort of university women. J Infect Dis 1996;174:927± 36 Olsen AO, Dillner J, Gjien K, Magnus P. Seropositivity against HPV 16 capsids: a better marker of past sexual behavior than presence of HPV DNA. Genitourin Med 1997; 73:131± 5 Schairer C, Brinton LA, Devesa SS, Ziegler RG, Fraumeni JF Jr. Racial differences in the risk of invasive squamous-cell cervical cancer. Cancer Causes and Control 1991;2:283± 9 Hildesheim A, Gravitt P, Schiffman MH, et al. Determinants of genital human papillomavirus infection in low-income women in Washington DC. Sex Transm Dis 1993;20:279± 85 Olsen AO, Dillner J, Gjoen K, Sauer T, Orstavik I, Magnus P. A population-based case± control study of human papillomavirus-type-16 seropositivity and incident high-grade dysplasia of the uterine cervix. Int J Cancer 1996;68:415± 19 Wideroff L, Schiffman M, Haderer P, et al. Seroreactivity to human papillomavirus types 16, 18, 31 and 45 virus-like particles in a case± control study of cervical squamous intraepithelial lesions. J Infect Dis 1999;180:1424± 8 Schiffman M, Herrero R, Hildesheim A, et al. HPV DNA testing in cervical cancer screening: results from women in a high-risk province of Costa Rica. JAMA 2000;283:87± 93

(Accepted 10 March 2002)